SGI workstations
Loki Jorgenson Rm421
loki at physics.mcgill.ca
Mon Dec 3 03:42:40 AEST 1990
An enthusiast writes:
> Date: 1 Dec 90 22:45:56 GMT
> From: Dan Severance <cs.yale.edu!doctor.chem.yale.edu at cs.yale.edu>
> Organization: Laboratory for Computational Chemistry, Yale University
> Subject: I would like to see a low-end desktop model!!
> In short, it would be great if SGI came out with a sparcstation-like
>product, color, 8bp 2D graphics, but with the crunch-power of their 4D-25
>or 4D-35 machines. (our code runs on a 4D-25 2-2.5 times as fast than a
>SparcStation 1 and it likes larger cache a lot which the 4D-35 will have).
> If SGI had a machine like that, we wouldn't buy servers, we'd buy it
>instead so we'd have a workstation environment consistent with the 3D
>machines, but it would be plenty for editing, 2D graphics, etc...
... and I whole heartedly agree. We own several 4D/25's based
primarily on their strong graphics capabilities. We have an older SUN3
group as our base workstation machines. We use the speed/power of the
4D's extensively since they outstrip the SUNs easily. However, in
considering upgrading our system, we have two plans in mind:
1) Work farm: a single "server machine" with a monitor for user
interactive graphics and a big disk for data with a group of truly
"server" nodes (no monitors) to which users "farm out" their number-
crunching jobs. SGI or IBM is favoured for this job.
2) Workstation group: like our current SUN3's, a group of
high-powered workstations with a minimal windowing capability
for users to work on and do their program development and
prehaps 2D plotting. SUN Sparc 1/2 are favoured.
My point is, quite simply, that we are very happy with the IRIX
environment and don't see any need to expand to using the Sparcs and
SUN OS4.1 (we still use v3.5 on the SUN3's) EXCEPT.... the 4D/25s
are far too expensive with their super-whizzy graphics to be simple
workstations. Even the most basic 8-bit graphics is over-kill.
Consequently, we will have to split our time/money in an effort to meet
our needs. If SGI came out with very simple monitor and graphics which would
support the windowing system, we could avoid the complication of integrating
a new architecture into our group.
The "work farm" concept is extremely viable using SGI machines
because of the affordability of a unit without the graphics head. How
about an answer to the "workstation group" question, SGI?
Another enthusiast,
__ __
Loki Jorgenson / / \ \ node: loki at physics.mcgill.ca
Grad, Systems Manager / ////// \\\\\\ \ BITNET: PY29 at MCGILLA
Physics, McGill University \ \\\\\\ ////// / fax: (514) 398-3733
Montreal Quebec CANADA \_\ /_/ phone: (514) 398-6531
More information about the Comp.sys.sgi
mailing list